This is a kind of complementary to the book by Baker/Hacker on Language, Sense, and Nonsense. And this is quite funny, whereas it was a bit of a waste of time to read B/H because I knew beforehand that I would agree wholeheartedly I knew I would completely disagree with Katz. But he made me think and he provides me with very clear examples of what I consider to be rubbish. So it gets 3 stars.
The basic idea, or should one say the basic rhetorical trick, is to contrast a Democritean concept of language to a non-Democritean. Just like in physics where the mystery of appearance and reality disappears if one accepts a non-naive atomistic view of the universe so in linguistics "one could attain the idea that language has an underlying conceptual reality and instead drop the assumption that it is completely inaccessible." (p. 9)
"If linguists could discover the principles by which the speakers of o language perform the encoding and decoding of thoughts in communication, then they would succeed in doing what Wittgenstein said was impossible." (p. 16) And, of course, they did succeed.
Now, what is the big problem with language?
"The most remarkable fact about human speech communication is that [...] the sentences we speak and hear daily are new sentences that bear little or no physical resemblance to familiar ones." (I rather doubt that this is the case, but okay…) "Yet we understand almost every new sentence we encounter and our understanding is immediate. This is in striking contrast to our attempts at understanding new machines or gadgets […]" (p. 52)
One could just stop here. Because I came in contact with dozens of radios and I had never a problem of finding the know to switch a radio on and off or to adjust the volume. Sometimes it takes a little while to feel comfortable in a new car. But so far I always managed to drive. In contrast, I encountered many sentences, especially in philosophy that I could read and re-read and still made no sense.
So the basic premise of "tranformationalists" is just wrong (if you accept my anecdotal evidence.) I certainly do not think that I have an innate knowledge of radios. Because this is what this school of thought (coming from Chomsky) thinks is the solution to the mystery of our ability to understand "infinitely" many sentences.
(BTW I am always reminded of Douglas Adams who says somewhere that there is a theory that as soon as anyone discovers the meaning of the Universe, it is immediately replaced by something even more bizarre. So Chomsky started out by criticizing the bizarre behaviorist theory of Skinner and he certainly managed to replace it by something even more bizarre. To me the most amazing thing to happen in the history of human intellectual enterprise.)
Okay, so here is the "solution." A child learning his first language and a field linguist learning a foreign language are in much the same position, Katz says (which is quite wrong, see B&H). They "have to acquire a grammar". ("p. 139")
The difference is that the linguist uses learned overt knowledge and skills developed in training, whereas what the child does is done "on the basis of innate, covert capacities." "[...]the child chooses an optimal grammar from among a set of possible grammars determined by innate principles about the form of human language ... he is assumed to know, innately, that the grammar of the language has the form of a transformational grammar... Given this knowledge, the child can a priori construct the class of possible grammars, independently of any empirical data." (p.140)
"[...] the rationalist-transformational theory conceives of language learning as a process of verification and refutation of theories, much like the sort of process that goes on in the more advanced natural sciences." (p. 141)
For refutation, see B&H. But I think, it is not required.
There are a lot of other things one could mention. For example, the claim that usage of idioms (like "give hell to") are the exception. (p. 105) Or one could argue against his concept of analytical meaning. Maybe another time.